As usage increases, however, organizations begin to realize that AI is more than just a tool embedded in individual applications. The AI platform evolves into a permanently available service whose outputs actively shape workflows. In a growing number of companies, AI agents are autonomously taking over entire process steps. To do so, they require their own permissions to access data and resources.
The necessary shift in perspective is therefore clear: AI may start in Outlook, but its sphere of impact extends far beyond that. To effectively support the use of AI as a service, service management comes into play. It provides the service with a structured framework of specialized organizational capabilities and processes designed to optimize service quality and the customer experience across the entire lifecycle. This ensures that the service is aligned with both business and employee requirements and delivers the greatest possible value to the organization.
Maturity Assessment
Are you interested in learning more about how the use of M365 Copilot can be seamlessly integrated into your ITSM processes and tool landscape?
From Helper to Risk
In day-to-day operations, it quickly becomes apparent how demanding ongoing operations can be. Organizations face the following challenges:
AI platforms intervene in processes.
What initially feels like an assistive feature in Outlook, Teams, Word, Excel, or PowerPoint connects multiple data sources, follows role-based permissions, and interprets context.
This creates impulses that trigger decisions, set priorities, and initiate follow-on activities — up to and including processes in which prepared content is further processed.
Technology becomes either a backbone or a bottleneck.
The quality of AI-generated results depends heavily on permissions, data management, information availability, and integrations with third-party systems. If these factors are not maintained consistently, users perceive the effects as seemingly “unpredictable AI behavior.” This is especially true when service states are not transparently communicated and escalation or communication paths in the event of disruptions are unclear.
Support encounters new types of cases.
Requests suddenly sound like this:
- “Why is the AI platform delivering different results today?”
- “Why isn’t my ordering process working right now?”
- “Why is my colleague getting better results than I am?”
- “Why are incorrect or outdated details appearing here?”
These are rarely classic defects. Instead, the root causes typically lie in data maintenance, role and permission models, process configurations, or usage patterns. In addition, variable responses are inherent to AI. An AI model works with probabilities and recombines existing information, which means answers are never exactly the same. Support teams must understand this behavior and be able to explain it to users. In doing so, they play a key role in building trust and acceptance. Without service processes designed for this reality, tickets bounce back and forth between IT, business units, and security.
Without operational governance, every update becomes a risk.
AI platforms continuously evolve in production environments — new features, adjusted limitations, additional integrations. If operational governance is lacking to assess, approve, roll out, and communicate changes, updates reach the organization unprepared.
If there is no shared knowledge base capturing proven prompt patterns, common error scenarios, and secure usage practices — and keeping it up to date — each department ends up reinventing how to use the platform.
Responsibilities remain unclear.
Is the AI platform owned by IT, the business units, or security? Without a clearly defined service ownership model and established escalation paths, a responsibility vacuum emerges. Risks are identified but not addressed operationally; adjustments are made, but without structured impact monitoring.
From Risk to Control
With the increasing integration of AI mechanisms such as M365 Copilot, new operational requirements are emerging. Processes become more dependent on data, role models, and continuous development. IT Service Management (ITSM) provides the structured framework needed to manage this dynamic and make AI platforms reliably usable. It ensures that the required IT services are delivered effectively, in a standardized and demand-oriented manner, thereby offsetting the operational uncertainty that arises when deploying AI-enabled services.
An effective organization, supported by the right tools and end-to-end process orchestration, creates the stability required for AI-dependent workflows. As a result, fluctuations in technical behavior, shifting dependencies, or unclear responsibilities no longer pose a risk but instead become transparent and controllable. What matters most is integration into the existing service environment and the company’s value chains. Only then can the impact of AI platforms on business processes be reliably assessed and actively managed.
From an operational perspective, ITSM ensures that typical uncertainties in dealing with AI no longer arise in an unstructured way but are addressed systematically. Proactive monitoring allows irregularities to be identified early, before they affect process quality. Rapid incident resolution, detailed root cause analysis, and controlled implementation of changes prevent unexpected results, variable outputs, or technical deviations from spreading unchecked across the organization. At the same time, application availability increases and downtime is deliberately reduced.
In parallel, ITSM creates the necessary transparency around dependencies and risks. By documenting how components are interconnected, what impact changes can have, or where error chains may propagate, complex AI relationships become operationally manageable. What is often perceived in everyday work as difficult-to-explain behavior is thus placed on a clear, traceable foundation.
Another key aspect is the clear assignment of responsibilities. Clearly defined roles and established escalation paths enable faster decision-making and more targeted incident resolution. This is a critical factor when AI services are used concurrently across different areas and multiple underlying causes may be at play.
When ITSM is integrated early into AI application use cases and new operating models, it creates a holistic approach that leads to optimized processes, higher productivity, and increased overall efficiency. In this way, an innovative tool becomes a reliable, manageable component of enterprise operations.
Enabling Successful Integration
The use of AI platforms such as Copilot or Foundry within an organization—and thus the integration of these services into an existing IT and service organization—raises a number of fundamental questions:
- How does the AI platform fit into the existing IT and service organization as a service or application?
- Which functional and non-functional requirements apply in connection with the AI platform?
- Which processes need to be adapted?
- How is the technical and organizational integration implemented?
- How is the IT service operated reliably and continuously improved?
- How should the user experience and interaction with the services ideally be designed?
These questions can be addressed through a structured, ITSM-based approach—one that spans from initial needs assessment to continuous improvement and step by step integrates AI services into the organization’s existing operating model.
1. Analysis and Conception
The integration of new services begins with a joint assessment. Business units and IT collect and evaluate their requirements and consolidate them into a target operating model that describes how the service should be managed going forward. At the same time, existing ITSM processes, service management tools, and organizational structures are analyzed to assess maturity, efficiency, and effectiveness. Reviewing the available documentation provides a comprehensive picture of the current state. This combination of requirements analysis and process evaluation forms the basis for making well-founded decisions on how AI platforms can be integrated into the existing service landscape. A structured IT Service Management assessment can serve as a valuable starting point here.
2. Design Phase
Based on the analysis, the detailed design of the AI service is defined. This includes functional and non-functional requirements, as well as planning for architecture, interfaces, security aspects, and—where manageable—SLAs. It also includes defined processes for requesting, approving, and assigning licenses. Compliance requirements and contingency concepts are considered at this stage as well.
Because externally operated AI services evolve at a rapid pace, proactive change handling is essential: new features, discontinued functionalities, or modified licensing models must be identified early and communicated to the organization in a timely manner.
In addition, necessary quality assurance mechanisms are designed, such as reports to monitor license consumption, usage, and support. The integration of knowledge base articles and self-service options into ITSM tools is addressed at this stage as well.
Subsequently, the organizational and technical prerequisites are established. Relevant boards and committees may be expanded, permissions assigned, workflows configured, and required interfaces prepared. A pilot tests the service under real-world conditions. If agents or extensions are implemented as part of the rollout, formal testing and deployment strategies are also required. Using a defined test strategy, pilot users, suitable environments, and a clear deployment playbook, the service is validated for production readiness. Formal sign-off provides the basis for the go/no-go decision.
Service management becomes particularly critical when moving beyond the core AI service. If extensions to third-party systems are developed or agents for process automation are implemented, the organization assumes full responsibility for operation, maintenance, and further development. In this scenario, all classic ITSM disciplines apply in full.
3. Transition Phase
During the transition phase, the service is transferred into operational use. Final documentation is made available, and any required training is planned and delivered. All necessary configuration items are recorded. The defined processes are rolled out step by step across the relevant services and organizational units and stabilized until they are firmly embedded.
At the conclusion, the line organization receives a clear handover package, including documentation of all open items. Lessons learned workshops help capture key insights before the complete project documentation is handed over. Formal project acceptance confirms that the implementation is complete and that AI platforms such as Copilot or Foundry are fully integrated into the ITSM landscape.
4. Operations
A continuous improvement process ensures that user feedback, operational metrics, and new requirements flow into optimizations and further development of the AI landscape. Ongoing operations also include typical governance tasks. License management ensures that licenses are assigned, revoked, and—where possible—managed automatically in line with demand. Usage reporting provides transparency on which licenses are actively used and where capacity remains underutilized.
In addition, the high pace of innovation in AI services requires proactive scouting of new features and changes, as well as early communication across the organization before functionalities are modified or discontinued.
A strong AI platform requires strong IT Service Management
Without a clearly structured operating and service management model, AI services remain vulnerable to fluctuations, inconsistencies, and unclear responsibilities. Only when governance, processes, and technical foundations are properly aligned can AI services reliably realize their full potential. ITSM provides the foundation on which a stable, secure, and scalable AI service can be operated and managed.
Campana & Schott supports organizations in getting started with a structured maturity assessment. The focus is on seamlessly integrating the use of the AI platform into existing ITSM processes and tool landscapes in order to ensure governance, transparency, and long-term operational stability—while delivering the best possible user experience.
Would you like to learn how Campana & Schott can support you with a maturity assessment?
Authors
Categories
#AI Transformation
#AI Infrastructure
